[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.197.90.95. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Download PDF
Figure 1.
Association Among Bleeding Events, Treatment, and Patient Characteristics
Association Among Bleeding Events, Treatment, and Patient Characteristics

The hazard ratios (HRs) were estimated by Cox proportional hazards regression models with propensity score weighting. The number of other comorbidities has been calculated as the sum of previous history of acute myocardial infarction, Alzheimer disease, related disorders or senile dementia, anemia, asthma, benign prostatic hyperplasia, cataract, chronic obstructive pulmonary disease, congestive heart failure, depression, diabetes mellitus, ischemic heart disease, hip or pelvic fracture, glaucoma, hyperlipidemia, osteoporosis, rheumatoid arthritis or osteoarthritis, breast cancer, colorectal cancer, prostate cancer, lung cancer, and endometrial cancer. Nonsteroidal anti-inflammatory drugs (NSAIDs) include diclofenac, ibuprofen, naproxen, ketoprofen, fenoprofen, flurbiprofen, piroxicam, meloxicam, mefenamic acid, and indomethacin. Antiplatelet agents include aspirin, clopidogrel, prasugrel, dipyridamol, ticlopidine, and ticagrelor. The concurrent risk score was calculated with the use of an algorithm as described in the text; higher scores predict greater medical spending. CMS indicates Centers for Medicare & Medicaid Services; TIA, transient ischemic attack. Error bars indicate 95% CIs.

Figure 2.
Hazard Ratios (HRs) for Bleeding Events by Anatomical Site and Treatment
Hazard Ratios (HRs) for Bleeding Events by Anatomical Site and Treatment

The HRs were estimated by Cox proportional hazards regression models with propensity score weighting that controlled for age, sex, race, chronic kidney disease, history of stroke or transient ischemic attack, hypertension, acquired hypothyroidism, history of bleeding in the year before treatment initiation, history of hospitalization in the year before treatment initiation, use of nonsteroidal anti-inflammatory drugs, use of antiplatelet agents, number of other Centers for Medicare & Medicaid Services priority comorbidities, and annual concurrent risk score. NOS indicates not otherwise specified. Error bars indicate 95% CIs.

Figure 3.
Hazard Ratios (HRs) for Bleeding Events by Anatomical Site, Subgroup, and Treatment
Hazard Ratios (HRs) for Bleeding Events by Anatomical Site, Subgroup, and Treatment

The HRs were estimated by Cox proportional hazards regression models with propensity score weighting. The general model controlled for age, sex, race, chronic kidney disease (CKD), history of stroke or transient ischemic attack, hypertension, acquired hypothyroidism, number of other Centers for Medicare & Medicaid Services (CMS) priority comorbidities, history of bleeding in the year before treatment initiation, history of hospitalization in the year before treatment initiation, use of nonsteroidal anti-inflammatory drugs, use of antiplatelet agents, and annual concurrent risk score. Subgroup analyses controlled for the same covariates except for the one defining the subgroup. For example, age-stratified analysis controlled for the same covariates except for age. The number of comorbidities used to define the subgroup with 7 or more comorbidities was calculated as the sum of all CMS priority comorbidities but atrial fibrillation (previous history of acute myocardial infarction, Alzheimer disease, related disorders or senile dementia, anemia, asthma, benign prostatic hyperplasia, cataract, chronic obstructive pulmonary disease, congestive heart failure, depression, diabetes mellitus, ischemic heart disease, hip or pelvic fracture, glaucoma, hyperlipidemia, osteoporosis, rheumatoid arthritis or osteoarthritis, breast cancer, colorectal cancer, prostate cancer, lung cancer and endometrial cancer, history of stroke or transient ischemic attack, hypertension, acquired hypothyroidism, and CKD). Error bars indicate 95% CIs.

Table 1.  
Baseline Characteristics of the Cohorts, Before and After Propensity Score Weighting, by Treatment Group
Baseline Characteristics of the Cohorts, Before and After Propensity Score Weighting, by Treatment Group
Table 2.  
Adjusted Incidence Rates of Bleeding Events by Treatment Group
Adjusted Incidence Rates of Bleeding Events by Treatment Group
1.
Summary review for new drug application—dabigatran etexilate.2010.www.accessdata.fda.gov/drugsatfda_docs/nda/2010/022512Orig1s000SumR.pdf. Accessed October 3, 2013.
2.
Connolly  SJ, Ezekowitz  MD, Yusuf  S,  et al; RE-LY Steering Committee and Investigators.  Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361(12):1139-1151.
PubMedArticle
3.
Cano  EL, Miyares  MA.  Clinical challenges in a patient with dabigatran-induced fatal hemorrhage. Am J Geriatr Pharmacother. 2012;10(2):160-163.
PubMedArticle
4.
Chen  BC, Viny  AD, Garlich  FM,  et al.  Hemorrhagic complications associated with dabigatran use. Clin Toxicol (Phila). 2012;50(9):854-857.
PubMedArticle
5.
Béné  J, Saïd  W, Rannou  M, Deheul  S, Coupe  P, Gautier  S.  Rectal bleeding and hemostatic disorders induced by dabigatran etexilate in 2 elderly patients. Ann Pharmacother. 2012;46(6):e14.
PubMedArticle
6.
Wychowski  MK, Kouides  PA.  Dabigatran-induced gastrointestinal bleeding in an elderly patient with moderate renal impairment. Ann Pharmacother. 2012;46(4):e10.
PubMedArticle
7.
Thomas  K. A promising drug with a flaw. The New York Times. November 3, 2012:B1.
8.
Harper  P, Young  L, Merriman  E.  Bleeding risk with dabigatran in the frail elderly. N Engl J Med. 2012;366(9):864-866.
PubMedArticle
9.
Southworth  MR, Reichman  ME, Unger  EF.  Dabigatran and postmarketing reports of bleeding. N Engl J Med. 2013;368(14):1272-1274.
PubMedArticle
10.
Sipahi  I, Celik  S, Tozun  N.  A comparison of results of the US Food and Drug Administration’s Mini-Sentinel Program with randomized clinical trials: the case of gastrointestinal tract bleeding with dabigatran. JAMA Intern Med. 2014;174(1):150-151.
PubMedArticle
11.
Larsen  TB, Rasmussen  LH, Skjøth  F,  et al.  Efficacy and safety of dabigatran etexilate and warfarin in “real-world” patients with atrial fibrillation: a prospective nationwide cohort study. J Am Coll Cardiol. 2013;61(22):2264-2273.
PubMedArticle
12.
Riihimäki  J, Sund  R, Vehtari  A.  Analysing the length of care episode after hip fracture: a nonparametric and a parametric Bayesian approach. Health Care Manag Sci. 2010;13(2):170-181.
PubMedArticle
13.
Deitelzweig  SB, Pinsky  B, Buysman  E,  et al.  Bleeding as an outcome among patients with nonvalvular atrial fibrillation in a large managed care population. Clin Ther. 2013;35(10):1536, e1.
PubMedArticle
14.
Jasuja  GK, Reisman  JI, Miller  DR,  et al.  Identifying major hemorrhage with automated data: results of the Veterans Affairs Study to Improve Anticoagulation (VARIA). Thromb Res. 2013;131(1):31-36.
PubMedArticle
15.
Gage  BF, Waterman  AD, Shannon  W, Boechler  M, Rich  MW, Radford  MJ.  Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285(22):2864-2870.
PubMedArticle
16.
Hirano  K, Imbens  GW.  Estimation of causal effects using propensity score weighting: an application to data on right heart catheterization. Health Serv Outcomes Res Methodol. 2001;2(3-4):259-278.Article
17.
Kurth  T, Walker  AM, Glynn  RJ,  et al.  Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. Am J Epidemiol. 2006;163(3):262-270.
PubMedArticle
18.
Sato  T, Matsuyama  Y.  Marginal structural models as a tool for standardization. Epidemiology. 2003;14(6):680-686.
PubMedArticle
19.
Eikelboom  JW, Wallentin  L, Connolly  SJ,  et al.  Risk of bleeding with 2 doses of dabigatran compared with warfarin in older and younger patients with atrial fibrillation: an analysis of the Randomized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) trial. Circulation. 2011;123(21):2363-2372.
PubMedArticle
20.
Zhang  Y, Baicker  K, Newhouse  JP.  Geographic variation in Medicare drug spending. N Engl J Med. 2010;363(5):405-409.
PubMedArticle
21.
Zhang  Y, Wu  SH, Fendrick  AM, Baicker  K.  Variation in medication adherence in heart failure. JAMA Intern Med. 2013;173(6):468-470.
PubMedArticle
22.
Hart  RG, Boop  BS, Anderson  DC.  Oral anticoagulants and intracranial hemorrhage: facts and hypotheses. Stroke. 1995;26(8):1471-1477.
PubMedArticle
Original Investigation
January 2015

Risk of Bleeding With Dabigatran in Atrial Fibrillation

Author Affiliations
  • 1Department of Health Policy and Management, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania
  • 2Department of Obstetrics and Gynecology, La Paz University Hospital, Madrid, Spain
JAMA Intern Med. 2015;175(1):18-24. doi:10.1001/jamainternmed.2014.5398
Abstract

Importance  It remains unclear whether dabigatran etexilate mesylate is associated with higher risk of bleeding than warfarin sodium in real-world clinical practice.

Objective  To compare the risk of bleeding associated with dabigatran and warfarin using Medicare data.

Design, Setting, and Participants  In this retrospective cohort study, we used pharmacy and medical claims in 2010 to 2011 from a 5% random sample of Medicare beneficiaries. We identified participants as those newly diagnosed as having atrial fibrillation from October 1, 2010, through October 31, 2011, and who initiated dabigatran or warfarin treatment within 60 days of initial diagnosis. We followed up patients until discontinued use or switch of anticoagulants, death, or December 31, 2011.

Exposures  Dabigatran users (n = 1302) and warfarin users (n = 8102).

Main Outcomes and Measures  We identified any bleeding events and categorized them as major and minor bleeding by anatomical site. Major bleeding events included intracranial hemorrhage, hemoperitoneum, and inpatient or emergency department stays for hematuria, gastrointestinal, or other hemorrhage. We used a propensity score weighting mechanism to balance patient characteristics between 2 groups and Cox proportional hazards regression models to evaluate the risk of bleeding. We further examined the risk of bleeding for 4 subgroups of high-risk patients: those 75 years or older, African Americans, those with chronic kidney disease, and those with more than 7 concomitant comorbidities.

Results  Dabigatran was associated with a higher risk of bleeding relative to warfarin, with hazard ratios of 1.30 (95% CI, 1.20-1.41) for any bleeding event, 1.58 (95% CI, 1.36-1.83) for major bleeding, and 1.85 (95% CI, 1.64-2.07) for gastrointestinal bleeding. The risk of intracranial hemorrhage was higher among warfarin users, with a hazard ratio of 0.32 (95% CI, 0.20-0.50) for dabigatran compared with warfarin. Dabigatran was consistently associated with an increased risk of major bleeding and gastrointestinal hemorrhage for all subgroups analyzed. The risk of major bleeding among dabigatran users was especially high for African Americans and patients with chronic kidney disease.

Conclusions and Relevance  Dabigatran was associated with a higher incidence of major bleeding (regardless of the anatomical site), a higher risk of gastrointestinal bleeding, but a lower risk of intracranial hemorrhage. Thus, dabigatran should be prescribed with caution, especially among high-risk patients.

Introduction

The Food and Drug Administration (FDA) approved dabigatran etexilate mesylate for the prevention of stroke and systemic embolism in patients with nonvalvular atrial fibrillation (AF) in October 20101 based on the results of the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial.2Quiz Ref IDThe RE-LY trial did not find differences in the rates of major bleeding between treatments; however, dabigatran was superior to warfarin sodium in the prevention of stroke. In addition, dabigatran etexilate mesylate, 150 mg, was associated with a higher risk of gastrointestinal bleeding but a lower rate of intracranial hemorrhage compared with warfarin.

Several months after the approval of dabigatran, the FDA received a large number of reports of severe dabigatran-related bleeding reports through its Adverse Event Reporting System. Concurrently, several case reports of dabigatran-induced bleeding were published in medical journals36 and discussed in the mainstream media.7 The incidence of serious bleeding was especially high among elderly users and users with renal impairment.5,6,8

On the basis of a study9 that did not adjust for patient characteristics, the FDA concluded that the incidence of gastrointestinal and intracranial bleeding was not higher for dabigatran compared with warfarin. In contrast, a meta-analysis10 found that dabigatran was associated with a significantly higher risk of gastrointestinal bleeding. In addition, a 2013 study11 using data from Danish National Registries found that the rates of major and gastrointestinal bleeding for warfarin and dabigatran were similar.

It is still unclear whether the use of dabigatran leads to more bleeding compared with warfarin, especially in the real-world clinical practice. Using 2010 to 2011 Medicare Part D data, we examine this question among patients with AF.

Methods
Data Source and Study Population

The study was approved by the institutional review board at the University of Pittsburgh as exempt because existing deidentified data were used. We obtained a 5% random sample of Medicare beneficiaries in 2010 and 2011 (most recent data available) from the Centers for Medicare & Medicaid Services (CMS). We identified patients who were newly diagnosed as having AF from October 1, 2010, through October 31, 2011, by using the CMS Chronic Condition Warehouse indicator that traced the first diagnosis date back to January 1, 1999. The diagnosis of AF was defined as having 1 inpatient or 2 outpatient claims with primary or secondary International Classification of Diseases, Ninth Revision (ICD-9), code 427.31.12 We also required that individuals in our study sample had filled an outpatient prescription for either dabigatran or warfarin within 2 months of the first diagnosis (N = 9562). Those who filled prescriptions for dabigatran and warfarin during the first 2 months after diagnosis were excluded (N = 158). We followed up each individual from the first prescription of dabigatran or warfarin until discontinuation of use for more than 60 days, switch of anticoagulants, death, or December 31, 2011. Our final overall study sample included 1302 dabigatran users and 8102 warfarin users.

Outcomes

We identified 9 types of bleeding according to anatomical position: intracranial hemorrhage, hemoperitoneum, gastrointestinal bleeding, hematuria, epistaxis, hemoptysis, vaginal hemorrhage, hemarthrosis, and not otherwise specified (NOS) hemorrhage (the ICD-9 codes for these bleeding events are listed in eTable 1 in the Supplement).13 We categorized bleeding events as major and minor events. Major bleeding events included intracranial hemorrhage, hemoperitoneum, and inpatient or emergency department stays for gastrointestinal, hematuria, or NOS hemorrhage; minor bleeding events included epistaxis, hemoptysis, vaginal hemorrhage, hemarthrosis and any outpatient claim for hematuria, gastrointestinal, and NOS hemorrhage. Any bleeding included major and minor bleeding events. Several claims for the same type of bleeding made within 1 week were considered as the same event to avoid double counting.14

We defined time to bleeding as days between the first warfarin or dabigatran prescription and the date of the bleeding event. We analyzed the time to the first bleeding event, as well as the time to the first major hemorrhage, intracranial hemorrhage, gastrointestinal bleeding, hematuria, vaginal bleeding, hemarthrosis, hemoptysis, epistaxis, and NOS hemorrhage.

Covariates

We adjusted for 2 main categories of covariates: demographic variables and clinical characteristics. Demographic variables included age, sex, race, and Medicaid eligibility. Clinical characteristics included CHADS2 (congestive heart failure, hypertension, age of 75 years or older, diabetes mellitus, and stroke) score, chronic kidney disease, hypertension, history of stroke or transient ischemic attack, history of acute myocardial infarction, diabetes mellitus, congestive heart failure, acquired hypothyroidism, the number of other CMS priority comorbidities (Table 1), history of bleeding in the year before treatment initiation, history of hospitalization in the year before treatment initiation, use of nonsteroidal inflammatory drugs (NSAIDs), use of antiplatelet agents, and the CMS prescription drug hierarchical condition category (CMS-RxHCC) score. Quiz Ref IDThe CHADS2 score predicts the risk of stroke in patients with AF; in calculating CHADS2, a history of previous stroke or transient ischemic attack is assigned 2 points; congestive heart failure, hypertension, age of 75 years or older, and diabetes are each assigned 1 point. The score is calculated as the sum of all points.15 The ICD-9 codes used to identify the conditions mentioned are listed in eTable 2 in the Supplement. History of bleeding was defined as having one medical claim with ICD-9 codes for any bleeding in the year before treatment initiation, and history of any hospitalization was defined as having at least one inpatient admission related to any condition in the year before treatment initiation. We defined the use of NSAIDs as having at least one prescription for diclofenac, ibuprofen, naproxen, ketoprofen, fenoprofen calcium, flurbiprofen, piroxicam, meloxicam, mefenamic acid, or indomethacin after treatment initiation. Use of antiplatelet agents was defined as having at least one pharmacy claim for aspirin, clopidogrel bisulfate, prasugrel, dipyridamol, ticlopidine hydrochloride, and ticagrelor after treatment initiation. The CMS-RxHCC score was calculated using CMS Prescription Drug Hierarchical Condition Categories software that was downloaded from the CMS website based on the beneficiary demographic variable and medical diagnosis captured on the claims (http://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Risk-Adjustors-Items/Risk2006-2011.html). The CMS-RxHCC is used by the CMS to adjust payments for the Medicare Part D plans and therefore is a proxy for health status, with a higher value indicating worse health status.

Statistical Analysis

We compared patient demographic and clinical characteristics between dabigatran and warfarin users before and after propensity score weighting. Continuous variables were tested using the Fisher F test; categorical variables were compared using χ2 or Fisher exact test.

We constructed Cox proportional hazards regression models with propensity score weighting to examine the incidence of bleeding by severity and anatomical site. Propensity score weighting was used to balance patient characteristics between treatment groups and was conducted in 2 stages. In the first stage, we performed a multivariate logistic regression to predict the probability of being a dabigatran or warfarin user, controlling for all the listed covariates. In the second stage, we constructed Cox proportional hazards regression models to compare the hazard rates of bleeding between dabigatran and warfarin groups, using the inverse of the propensity score as a weight. This method assigned higher weights to individuals in one treatment group who had similar characteristics to those in the other group.1618 Each survival model also controlled for age, sex, race, chronic kidney disease, history of stroke or transient ischemic attack, hypertension, acquired hypothyroidism, number of other CMS priority comorbidities, history of bleeding, history of hospitalization, use of NSAIDs, use of antiplatelet agents, and annual concurrent risk score. Discontinuation of use of an anticoagulant (defined as a gap ≥60 days), switching to the other anticoagulant, death, and the end of the study period (December 31, 2011) were considered censoring events. Adjusted incidence rates for each outcome were calculated using the percentage of event-free individuals at the end of the study period, which was obtained from the Cox proportional hazards regression models.

We further examined the incidence of bleeding in subgroups stratified by age (<75 or ≥75 years) and among African Americans, users with renal impairment, and patients with at least 7 priority CMS conditions other than AF.12 Subgroup analyses were performed following the same methods and controlling for all covariates except for the one defining the subgroup. For example, age-stratified analysis controlled for the same covariates except for age.

Results
Patient Characteristics

The mean follow-up period in the overall sample was 177 days (interquartile range, 89-256 days) for dabigatran users and 228 days (interquartile range, 119-333 days) for warfarin users. Before propensity score weighting, some differences were found between dabigatran and warfarin users. For example, warfarin users were more likely to belong to minority racial groups, be eligible for Medicaid coverage, and have a history of bleeding and hospitalization (Table 1). The prevalence of chronic kidney disease, congestive heart failure, diabetes, and history of stroke or transient ischemic attack was significantly higher among the warfarin cohort. After propensity score weighting, however, the 2 groups were well balanced in all characteristics (Table 1).

Bleeding by Severity

Quiz Ref IDDabigatran was associated with a statistically significantly higher risk of major and any bleeding than warfarin after controlling for patient characteristics with inverse propensity score weighting. In particular, the adjusted incidence of major bleeding was 9.0% (95% CI, 7.8%-10.2%) for the dabigatran group and 5.9% (95% CI, 5.1%-6.6%) for the warfarin group (Table 2). Compared with warfarin, the hazard ratios (HRs) associated with dabigatran were 1.58 (95% CI, 1.36-1.83) for major bleeding and 1.30 (95% CI, 1.20-1.41) for any bleeding (Figure 1).

In addition, some demographic and clinical factors were associated with a higher risk of bleeding for both treatment groups. Age of 75 years or older increased the hazard rate of major bleeding (HR, 1.48; 95% CI, 1.21-1.96) compared with those younger than 65 years. Blacks had a higher likelihood of major and any bleeding, with HRs of 2.09 (95% CI, 1.68-2.60) for major bleeding and 1.16 (95% CI, 1.01-1.34) for any bleeding relative to whites. Chronic kidney disease and antiplatelet use were associated with a higher hazard rate of major and any bleeding.

Bleeding by Anatomical Site

Relative to warfarin, dabigatran was associated with higher HRs of 5 bleeding events: 1.85 (95% CI, 1.64-2.07) for gastrointestinal bleeding, 1.41 (95% CI, 1.21-1.64) for hematuria, 2.27 (95% CI, 1.32-3.90) for vaginal bleeding, 2.78 (95% CI, 1.32-5.86) for hemarthrosis, and 1.49 (95% CI, 1.04-2.14) for hemoptysis (Figure 2). However, dabigatran users had a lower adjusted rate of intracranial bleeding of 0.6% (95% CI, 0.3%-0.8%) compared with a rate of 1.8% (95% CI, 1.4%-2.2%) among warfarin users (Table 2), with an HR of 0.32 (95% CI, 0.20-0.50) for dabigatran compared with warfarin (Figure 2). The incidence of epistaxis and nonspecified hemorrhage was also lower among dabigatran users; dabigatran was associated with HRs of 0.65 (95% CI, 0.49-0.85) for epistaxis and 0.74 (95% CI, 0.61-0.90) for nonspecified hemorrhage relative to warfarin.

Subgroup Analyses

The unadjusted rates of bleeding events by subgroups are given in eTable 3 in the Supplement. After adjusting for patient characteristics, dabigatran was associated with an increased risk of major or any bleeding events and gastrointestinal hemorrhage for all subgroups analyzed (Figure 3). The risk of major bleeding among dabigatran users was especially high for blacks (HR, 2.12; 95% CI, 1.39-3.24) and patients with chronic kidney disease (HR, 2.07; 95% CI, 1.66-2.58), both relative to warfarin. The age-stratified results for intracranial bleeding indicated that warfarin increased the risk of intracranial hemorrhage for patients older than 75 years, with an HR of 0.10 (95% CI, 0.04-0.24) for dabigatran compared with warfarin. However, the hazard rates of intracranial bleeding for patients younger than 75 years and African Americans were not different between the treatment groups.

Discussion

To the best of our knowledge, our study is the first to compare the safety profile of dabigatran and warfarin using a nationally representative sample of Medicare beneficiaries. Quiz Ref IDWe found that in the real-world clinical practice, after adjusting for patient clinical and demographic characteristics, dabigatran was associated with a higher incidence of major bleeding regardless of the anatomical site; in addition, dabigatran was associated with a higher risk of gastrointestinal bleeding but a lower risk of intracranial hemorrhage than warfarin.

Our results for gastrointestinal and intracranial bleeding are consistent with the RE-LY trial. In addition, the RE-LY trial found that dabigatran was associated with a lower risk of major bleeding than warfarin among patients younger than 75 years but an increased risk for users older than 75 years19; however, we found that dabigatran was associated with a higher risk of major bleeding for both age groups. We found opposite results compared with the FDA’s investigation, which did not adjust for differences in patient characteristics between the treatment groups.9 Dabigatran and warfarin users are very different in several factors that directly affect the risk of bleeding, and failing to adjust would bias the results, as our unadjusted estimates indicate.

Quiz Ref IDOur findings are subject to 3 limitations. First, the propensity score weighting did not control for unobserved factors that may affect the initiation of warfarin or dabigatran. For example, patients with renal impairment who are taking warfarin may be more likely to have lower levels of creatinine clearance than patients with chronic kidney disease who are taking dabigatran. Because renal function is associated with bleeding, it would lead to a downward bias. However, we controlled for history of chronic kidney disease in the propensity score calculation and in the survival models. We also evaluated bleeding outcomes for patients with renal impairment as a subgroup. Second, our study focuses on evaluating the safety profile of both drugs. We only had a maximum 14-month follow-up period because 2011 Part D data are the most recently available data to researchers. We therefore could not evaluate the incidence of stroke, so our results cannot compare the benefit-risk ratio of treatments. However, we observed significant differences in the primary outcomes of our study. Third, as in other studies20,21 that use claims data, we lacked detailed laboratory results, such as creatinine clearance; therefore, we were not able to calculate the proportion of bleeding events because of failure to adjust the dabigatran dose for renal function.

Nevertheless, our study has several important clinical and policy implications. First, physicians should prescribe dabigatran with caution, especially among African Americans and patients with renal impairment. Second, the risk of gastrointestinal bleeding was consistently higher for all subgroups of dabigatran users compared with warfarin users; therefore, it is of special concern from the clinical perspective. As a result, it is important for physicians to explain to patients how to detect gastrointestinal bleeding so that it can be controlled as early as possible. Third, intracranial hemorrhage is the most feared complication associated with warfarin22; thus, patients at high risk of intracranial hemorrhage may be willing to accept the higher risk of other bleeding events associated with dabigatran for a lower likelihood of intracranial bleeding. Arguably, this is the subgroup in which dabigatran is most likely to be a favorable choice in terms of safety.

Conclusions

We found that dabigatran was associated with a higher incidence of major bleeding, a higher risk of gastrointestinal bleeding, but a lower risk of intracranial hemorrhage than warfarin. Before more evidence is available, dabigatran should be prescribed with caution in high-risk patients. Our study also indicates the importance of comparing the safety profiles of newer and traditional drugs using postmarketing real-world data. As newer Part D data become available, similar analyses could be extended for other new oral anticoagulants, such as rivaroxaban, apixaban, and edoxaban.

Back to top
Article Information

Accepted for Publication: July 23, 2014.

Corresponding Author: Yuting Zhang, PhD, Department of Health Policy and Management, University of Pittsburgh, 130 De Soto St, Crabtree Hall, Room A664, Pittsburgh, PA 15261 (ytzhang@pitt.edu).

Published Online: November 3, 2014. doi:10.1001/jamainternmed.2014.5398.

Author Contributions: Dr Hernandez had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Hernandez, Baik, Zhang.

Acquisition, analysis, or interpretation of the data: All authors.

Drafting of the manuscript: Hernandez, Zhang.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Hernandez, Baik.

Obtained funding: Zhang.

Administrative, technical, or material support: Piñera, Zhang.

Study supervision: Baik, Piñera, Zhang.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by grant R01 HS018657 from the Commonwealth Foundation and Agency for Healthcare Research and Quality (Dr Zhang).

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.

References
1.
Summary review for new drug application—dabigatran etexilate.2010.www.accessdata.fda.gov/drugsatfda_docs/nda/2010/022512Orig1s000SumR.pdf. Accessed October 3, 2013.
2.
Connolly  SJ, Ezekowitz  MD, Yusuf  S,  et al; RE-LY Steering Committee and Investigators.  Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361(12):1139-1151.
PubMedArticle
3.
Cano  EL, Miyares  MA.  Clinical challenges in a patient with dabigatran-induced fatal hemorrhage. Am J Geriatr Pharmacother. 2012;10(2):160-163.
PubMedArticle
4.
Chen  BC, Viny  AD, Garlich  FM,  et al.  Hemorrhagic complications associated with dabigatran use. Clin Toxicol (Phila). 2012;50(9):854-857.
PubMedArticle
5.
Béné  J, Saïd  W, Rannou  M, Deheul  S, Coupe  P, Gautier  S.  Rectal bleeding and hemostatic disorders induced by dabigatran etexilate in 2 elderly patients. Ann Pharmacother. 2012;46(6):e14.
PubMedArticle
6.
Wychowski  MK, Kouides  PA.  Dabigatran-induced gastrointestinal bleeding in an elderly patient with moderate renal impairment. Ann Pharmacother. 2012;46(4):e10.
PubMedArticle
7.
Thomas  K. A promising drug with a flaw. The New York Times. November 3, 2012:B1.
8.
Harper  P, Young  L, Merriman  E.  Bleeding risk with dabigatran in the frail elderly. N Engl J Med. 2012;366(9):864-866.
PubMedArticle
9.
Southworth  MR, Reichman  ME, Unger  EF.  Dabigatran and postmarketing reports of bleeding. N Engl J Med. 2013;368(14):1272-1274.
PubMedArticle
10.
Sipahi  I, Celik  S, Tozun  N.  A comparison of results of the US Food and Drug Administration’s Mini-Sentinel Program with randomized clinical trials: the case of gastrointestinal tract bleeding with dabigatran. JAMA Intern Med. 2014;174(1):150-151.
PubMedArticle
11.
Larsen  TB, Rasmussen  LH, Skjøth  F,  et al.  Efficacy and safety of dabigatran etexilate and warfarin in “real-world” patients with atrial fibrillation: a prospective nationwide cohort study. J Am Coll Cardiol. 2013;61(22):2264-2273.
PubMedArticle
12.
Riihimäki  J, Sund  R, Vehtari  A.  Analysing the length of care episode after hip fracture: a nonparametric and a parametric Bayesian approach. Health Care Manag Sci. 2010;13(2):170-181.
PubMedArticle
13.
Deitelzweig  SB, Pinsky  B, Buysman  E,  et al.  Bleeding as an outcome among patients with nonvalvular atrial fibrillation in a large managed care population. Clin Ther. 2013;35(10):1536, e1.
PubMedArticle
14.
Jasuja  GK, Reisman  JI, Miller  DR,  et al.  Identifying major hemorrhage with automated data: results of the Veterans Affairs Study to Improve Anticoagulation (VARIA). Thromb Res. 2013;131(1):31-36.
PubMedArticle
15.
Gage  BF, Waterman  AD, Shannon  W, Boechler  M, Rich  MW, Radford  MJ.  Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285(22):2864-2870.
PubMedArticle
16.
Hirano  K, Imbens  GW.  Estimation of causal effects using propensity score weighting: an application to data on right heart catheterization. Health Serv Outcomes Res Methodol. 2001;2(3-4):259-278.Article
17.
Kurth  T, Walker  AM, Glynn  RJ,  et al.  Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. Am J Epidemiol. 2006;163(3):262-270.
PubMedArticle
18.
Sato  T, Matsuyama  Y.  Marginal structural models as a tool for standardization. Epidemiology. 2003;14(6):680-686.
PubMedArticle
19.
Eikelboom  JW, Wallentin  L, Connolly  SJ,  et al.  Risk of bleeding with 2 doses of dabigatran compared with warfarin in older and younger patients with atrial fibrillation: an analysis of the Randomized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) trial. Circulation. 2011;123(21):2363-2372.
PubMedArticle
20.
Zhang  Y, Baicker  K, Newhouse  JP.  Geographic variation in Medicare drug spending. N Engl J Med. 2010;363(5):405-409.
PubMedArticle
21.
Zhang  Y, Wu  SH, Fendrick  AM, Baicker  K.  Variation in medication adherence in heart failure. JAMA Intern Med. 2013;173(6):468-470.
PubMedArticle
22.
Hart  RG, Boop  BS, Anderson  DC.  Oral anticoagulants and intracranial hemorrhage: facts and hypotheses. Stroke. 1995;26(8):1471-1477.
PubMedArticle
×